Patient population:

In this retrospective study, all 994 consecutive lung biopsies performed between January 2013 and December 2017 in our institution, a tertiary hospital located in Sao Paulo, Brazil, were retrieved from hospital records.  All samples were collected by intra-operative biopsy, bronchoscopy with transbronchial biopsy or percutaneous CT-guided needle biopsy.

This study was approved by institutional and national ethics committees. Due to the retrospective nature of the study, the informed consent was not necessary.

The inclusion criteria was the existence of granulomatous lesions in the pathology report (88 of 994 patients). Patients without chest CT previous to biopsy were excluded and the final study population was 75 patients (Fig. 1).

Data acquisition:

Demographic and pathologic data of all cases were reviewed and collected (C.V.O.). Two radiologists (each one with 1 year experience in chest radiology), blinded to the etiology of the GLD, reviewed the Chest CT of all patients and classified the imaging features of each case (D.R. and M.B.R.). A third radiologist (chest specialist with 10 years experience) reviewed all cases of disagreement to reach a consensus (H.J.L.).

Data analysis:

The CT findings were categorized by the main radiological pattern: consolidation, ground-glass opacity, mass, nodular or micronodular. Micronodular pattern was subcategorized into random, perilymphatic or centrilobular. Other analyzed CT features included: associated findings (cavity, mediastinal/hilar lymphadenopathy, pleural effusion), distribution (focal, multifocal or diffuse), location (right, left or both lungs) and specific affected pulmonary lobes.

Statistical analysis:

Categorical data were represented by absolute (n) and relative (%) frequency, and contingency matrices were analyzed by Pearson's chi-squared test. A risk α ≤ 5% for type I error and risk β ≤ 20% for type II error was considered for the present study. P-value of less than 0.05 was considered statistically significant.